Intelligent bidding method and system based on dynamic benchmark price

Abstract

The invention provides an intelligent bidding method and system based on dynamic benchmark price. The method comprises the steps of calculating a dynamic standard price after receiving a quoted price of a supplier; applying a stability priority decision rule based on the price; if the standard subject matter is the standardized subject matter, sorting the decisions according to the absolute value of the difference between the quoted price and the reference price; and if the subject matter is a non-standardized subject matter, a decision mark is sorted according to a comprehensive score, and the score is obtained by calculating a price sub-score according to a standard price and weighting the price sub-score with the credit and performance ability sub-scores. The system is correspondingly provided with an adaptive decision-making module, a credit portrait module, an information isolation module, an anti-serial connection module and a block chain evidence storage module. According to the invention, through cooperation of the dynamic benchmark price and the differentiation decision rule, the technical problems of poor system stability, serial bidding and high admission barrier of new suppliers caused by the rule of'winning bid at the lowest price 'are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, e-commerce, supply chain management and financial technology, in particular to an intelligent bidding method and system for realizing system stability and transaction quality optimization through dynamic benchmark price, adaptive algorithm and bidirectional ecological incentive. BACKGROUND

[0002] With the popularization of online bidding mode, network bidding such as reverse auction has become the core means of B2B, B2C and other transactions. The existing technology generally adopts the rules of "lowest price winning" or "calculating comprehensive score based on the lowest price as the benchmark". The original intention of the design of this rule is to stimulate the extreme price competition among suppliers, so as to realize cost saving for the purchaser.

[0003] However, the present inventors have found through long-term commercial practice and technical analysis that this bidding mechanism with "lowest price" as the single guide has inherent and not fully recognized systemic defects and design flaws, which leads the entire transaction ecosystem into an unsustainable "vicious cycle":

[0004] 1. System stability and quality risk: extreme price pressure seriously compresses the profit space of suppliers, triggering the "inferior coin drives out good coin" effect. In order to survive, some suppliers are forced to cut corners, leading to quality decline, or choose to default when there is no profit, resulting in order loss. Internal data of a large B2B platform shows that under the lowest price benchmark rule, the subsequent dispute rate and default rate of the winning order is 2-3 times the average level of the platform.

[0005] 2. High-quality supplier squeeze-out effect: high-quality suppliers who focus on long-term reputation and product quality gradually withdraw from such bidding due to their inability to withstand vicious price wars, leading to the degradation of the platform ecosystem and the long-term interests of the purchaser.

[0006] 3. Collusion behavior breeding ground: the lowest price model makes the winning result extremely sensitive to the slight change of the bid, which provides a clear motivation and simple operation model for suppliers to manipulate the winning order through coordinated bidding (i.e. collusion), seriously damaging the fairness of bidding.

[0007] 4. New supplier access barrier: for new suppliers, due to the lack of historical transaction data (such as compliance rate) in the platform, they are at a disadvantage in the comprehensive scoring system based on historical data, making it difficult to obtain a fair competition opportunity, forming an insurmountable access barrier.

[0008] Currently, the industry generally attributes the above problems to the integrity of suppliers, and has not recognized the harm from the root cause of the system rule design, nor has it proposed a technical solution that can fundamentally solve the problem.

[0009] Therefore, there is an urgent need in the art for an innovative bidding decision-making method that can abandon the traditional paradigm of simply pursuing low prices, while producing a reasonable price reduction effect, fundamentally repairing the system stability crisis caused by rule defects, and maintaining a healthy, stable, and sustainable online transaction ecosystem. SUMMARY

[0010] OBJECTIVE

[0011] The present application aims to overcome the above-mentioned inherent defects of the prior art and provide a new intelligent bidding method and system. The core objective is not simply to minimize the contract price number, but to achieve a paradigm shift from "price optimization" to "system stability and overall cost optimization" by introducing new decision-making rules and adaptive mechanisms, thereby fundamentally solving the systemic technical problems of insufficient system stability, declining transaction quality, strong collusion motivation, and difficulty in admitting new suppliers caused by rule defects.

[0012] The "system stability" referred to in the present application specifically refers to the output quality of the bidding process executed by the data processing system, and the key technical indicators include: final compliance rate, transaction dispute rate, order loss rate, and high-quality supplier retention rate.

[0013] Core principle: achieve the dual-closed-loop coupling mechanism of "high quality and low price"

[0014] The core principle of the present application to achieve "high quality and low price" is to construct two mutually coupled and synergistic automated closed-loop control processes: one is the "dynamic bidding cycle" that drives price discovery and optimization, and the other is the "quality assurance closed loop" that guarantees the output quality of the system.

[0015] Dynamic bidding cycle: triggered by the stability priority decision rule and the intelligent guidance step. The specific process is: the system calculates the dynamic benchmark price based on all bids, and sends personalized fuzzy guidance information to the suppliers whose bids are judged to deviate from the reasonable interval; the suppliers receiving the guidance information strategically adjust their bids to improve their winning probability, which directly leads to a decrease in the next round of dynamic benchmark price; this process continues to iterate within the bidding cycle, driving the overall bid level to converge towards a more market-competitive direction, thereby automatically achieving the "low price" goal.

[0016] The calculation of the dynamic benchmark price adopts an adaptive algorithm, the core of which is a small sample fault-tolerant mechanism: when the number of valid bids is greater than or equal to a first threshold value, the trimmed mean is used preferentially; when the number of bids is less than the first threshold value but greater than or equal to a second threshold value, the median is used; and when the number of bids is less than the second threshold value, the arithmetic mean is used and a weight adjustment strategy is triggered. The first threshold value and the second threshold value are configurable parameters, wherein the first threshold value is greater than the second threshold value. As a preferred embodiment, the first threshold value can be set to 5 and the second threshold value can be set to 3.

[0017] Quality assurance closed loop: then composed by the stability priority decision rule and the credit image module and the comprehensive score calculation. The specific process is: in the competitive bid set generated by the dynamic bidding cycle, which has competitive bids, the system assesses the reliability of the suppliers through the credit image module, and calculates the comprehensive score accordingly; the score system ensures that at the final bid, even if the bid of a supplier with poor credit record or insufficient performance ability has advantages, it will be eliminated by high-quality suppliers with excellent credit and reliable performance in the comprehensive score ranking, thereby systematically ensuring the output of "high quality".

[0018] The above two processes are deeply coupled and indispensable. If only the dynamic bidding cycle, the system will repeat the "inferior coin expelling good coin" pattern; if only the quality assurance closed loop, there is a lack of internal driving force for price competition. The present application successfully solves the technical paradox that "high quality" and "low price" cannot be achieved in the prior art by organically combining the two, thereby stably outputting "high quality and low price" in a real sense at the system level.

[0019] Technical scheme

[0020] To achieve the above object, the present application adopts the following technical scheme:

[0021] In a first aspect, the present application provides an intelligent bidding method based on a dynamic benchmark price, which is executed by one or more processors of a data processing system, comprising the following steps:

[0022] Receiving bids from multiple suppliers for a specific target, the target including a dynamic order pool aggregated by transaction requests of multiple demand parties;

[0023] Calculating a dynamic benchmark price of all valid bids;

[0024] Based on the dynamic benchmark price, applying a stability priority decision rule to determine the winning bidder;

[0025] Wherein, the stability priority decision rule is:

[0026] According to the attribute information of the target object, it is judged whether it is a standardized target object; if it is a standardized target object, the absolute values of the differences between the supply side offer prices and the dynamic benchmark price are sorted from small to large, and the best bid is determined;

[0027] If it is a non-standardized target object, it is sorted and determined according to the comprehensive score from high to low, and the comprehensive score is at least calculated by weighting the price sub-score, the credit sub-score and the performance sub-score, and the price sub-score is calculated based on the dynamic benchmark price.

[0028] The performance sub-score is calculated by the following formula:

[0029] Performance sub-score = [W1×(actual completed order number / total accepted order number)+W2×(1- average delay delivery days / average agreed delivery days)+W3×(1-quality defect order number / total performance order number)]×100×W p

[0030] Wherein:

[0031] W1, W2, W3 are internal weight coefficients of performance rate, timeliness of performance and quality of performance respectively, and W1+W2+W3=1.

[0032] W p is the total weight coefficient of the performance sub-score in the comprehensive score, which is coordinated with the weights of the price and credit sub-scores, and the sum of the three is 1.

[0033] For a new supplier, the system directly gives it a preset performance capability basic score (such as 80 points). For a supplier with less than 10 historical orders, the variables in the formula are calculated based on its completed orders only.

[0034] The calculation formula of the price sub-score is: price score = MAX(0,(1- |offer-dynamic benchmark price| / dynamic benchmark price))×100×price weight coefficient.

[0035] The price weight coefficient is a dynamic parameter, and its value is dynamically determined by the system according to at least one of the following factors: the standardization degree of the target object, the real-time market supply and demand index, and the credit rating level of the supplier.

[0036] The real-time market supply and demand index is used to quantify the real-time competition situation of the market, which is calculated based on the ratio of the total supply capacity of the supplier to the total demand of the demand side. As a preferred embodiment, it is specifically defined as: the ratio of the total supply of the participating supplier to the aggregated order demand of the demand side.

[0037] The dynamic adjustment strategy of the price weight coefficient includes but is not limited to the following embodiments:

[0038] Embodiment A: When the real-time market supply-demand index is higher than a preset first threshold value, indicating that the market is in a state of oversupply, the system automatically reduces the price weight coefficient. As a preferred configuration, the first threshold value can be in the range of 1.0 to 1.5, for example, set to 1.2. The reduction range can be set in the range of 15% to 25% of the base value, for example, preferably reduced by 20%.

[0039] Embodiment B: When the credit score of the supplier is lower than a preset second threshold value, indicating that there is a risk of performance reliability, the system automatically reduces the price weight coefficient in the calculation of the comprehensive score of the supplier. As a preferred configuration, the second threshold value can be in the range of 75 to 85, for example, set to 80. The additional reduction range can be set in the range of 5% to 15%, for example, preferably reduced by 10% additionally.

[0040] The skilled in the art can understand that the above-mentioned values are only preferred examples, and the actual configuration can be based on the scene, and any transformation based on this core idea falls within the protection scope of the present application.

[0041] The method further comprises an intelligent guidance step: the system generates personalized and asymmetric fuzzy guidance information for each supplier based on the dynamically calculated benchmark price. The guidance information does not contain any precise aggregate data that can be used as collusion benchmark; wherein the guidance strategy is asymmetric, the system only sends quantitative guidance information to suppliers whose prices are higher than the reasonable interval, and does not send or only sends neutral prompt information to suppliers whose prices are too low or in the reasonable interval. The reasonable interval is a price range determined based on the dynamic benchmark price, with a system preset percentage of up and down. The typical value range of the preset percentage is 5% to 15%, which can be set to default value according to the type of target (such as 10% for standardized goods and 15% for non-standard services).

[0042] The method further comprises an active anti-collusion step:

[0043] Complete stealth and fuzzy guidance: the system does not provide any form of participant identification or public benchmark information in the foreground.

[0044] Background behavior tracking and anonymous identification: a random anonymous identification is generated and maintained for each supplier in the background of the system, which is associated with the real identity of the supplier and is periodically reset.

[0045] When a tentative bidding pattern is detected, the random anonymous identifier is immediately reset non-periodically. The tentative bidding pattern refers to a characteristic sequence identified by the system's backend time-series behavior analysis model, where at least two suppliers maintain a stable bid difference within ±1% over two or more consecutive bidding rounds, and the bid time intervals exhibit a fixed pattern. (The "±1%" is a preferred value and can be adjusted according to actual scenario requirements.)

[0046] A time-series behavior analysis model is used to identify complex collusion patterns, and the identified abnormal behaviors are processed in a hierarchical manner.

[0047] The method also includes information isolation and rule-based evidence preservation steps:

[0048] A) Information Isolation: Implement a role-based, global zero-knowledge information isolation strategy:

[0049] For suppliers, the identity and pricing information of all other suppliers, as well as the identity information of the demanders, are completely hidden.

[0050] For the demand side, the identity information of all other demanders, as well as the identity information and pricing information of all suppliers, are completely hidden.

[0051] For non-participants and tourists, only highly abstract and aggregated de-identified information is provided.

[0052] For system administrators, implement a "logical isolation and on-chain evidence storage" mechanism to prevent internal data leakage.

[0053] B) Rule Preservation: Key data throughout the bidding process (including but not limited to bidding decision rules, all valid bids, dynamic benchmark price, and winning bid results) will generate hash values ​​and be written to the blockchain. Preferably, the blockchain adopts a consortium blockchain architecture to ensure data confidentiality during the bidding process and verifiability after the bid is awarded. The consortium blockchain is jointly maintained by this bidding platform, a judicial appraisal center, and industry associations. Only hash values ​​are written during the bidding process, and data decryption or public disclosure for verification is authorized only after the bid is awarded. This enhances the platform's credibility.

[0054] Secondly, the present invention provides an intelligent bidding system for implementing the above method, which adopts a modular architecture design and includes a processor, a storage medium, and a core functional module running on the processor. The core functional module includes:

[0055] The adaptive decision engine module is configured to perform dynamic benchmark price calculation, adaptive weight coefficient adjustment, and intelligent guidance strategy generation.

[0056] The credit profiling and ecosystem management module is configured to integrate multi-source data to generate credit assessments for suppliers and demanders, and to assign a preset credit score base to new suppliers that lack historical data.

[0057] The strict information isolation control module is configured to implement a role-based, global zero-knowledge information isolation strategy;

[0058] The proactive anti-collusion module is configured to perform dynamic anonymous identifier management, collusion intent identification, and behavior sequence analysis, judgment, and handling.

[0059] The blockchain evidence storage and smart contract module is configured to handle the entire process of data on-chain, deploy and manage smart contracts, and initiate an AI-assisted on-chain arbitration process in the event of a dispute.

[0060] The core functional modules of the system communicate and call each other through predefined interfaces; these interfaces are configured to connect to external data sources and extension plugins that conform to their specifications, so as to realize the system's functional expansion, technical upgrades and update maintenance.

[0061] Beneficial effects

[0062] Compared with the closest existing technology, this invention, through systematic rule reconstruction and algorithm optimization, brings about the following unexpected and significant technical advancements:

[0063] The system has achieved a qualitative leap in stability and transaction quality: by avoiding the "lowest price" rule that induces vicious competition and introducing a robust benchmark price and a rigid credit threshold, the reasonable profit margin and performance capability of the suppliers are fundamentally guaranteed, thereby significantly improving the final performance rate of the system and keeping the transaction dispute rate at an extremely low level.

[0064] Achieving true "high quality at a low price" and optimal total cost: This invention, through the coupling of a "dynamic bidding cycle" and a "quality assurance closed loop," abandons unsustainable "bloody low prices" and instead outputs reliable and sustainable "fair low prices." Although the average transaction price reduction may not be the largest, thanks to its extremely high fulfillment rate and near-zero dispute rate, the technical solution of this invention can achieve absolute minimization of the total cost of ownership for the demand side.

[0065] Collusion motives and behaviors are systematically suppressed and actively combated: Due to the adoption of multiple mechanisms combining role-based full-domain zero-knowledge information isolation, personalized fuzzy guidance, backend dynamic behavior analysis, and on-chain evidence storage of administrator operations, the possibility of collusion is completely eradicated at the information level.

[0066] Building a healthy and sustainable platform technology ecosystem: By using a two-way credit system, a new supplier credit scoring mechanism, and a small-sample fault-tolerance strategy, we have broken down entry barriers through technological means and created a level playing field.

[0067] Providing judicial-grade credible evidence and enabling intelligent and automated operation and maintenance: Blockchain-based end-to-end evidence storage provides an immutable and credible chain of evidence for dispute resolution. By introducing AI as an arbitration aid, the efficiency and analytical depth of the arbitration process are improved; finally, combined with the automatic execution of smart contracts, the entire chain from evidence preservation and analytical assistance to award enforcement is automated. Attached Figure Description

[0068] Figure 1 This is a schematic diagram of the modular architecture of the intelligent bidding system of the present invention.

[0069] Figure 2 This is the overall flowchart of the intelligent bidding method of the present invention.

[0070] Figure 3 This is a flowchart of the comprehensive score calculation and the adaptive calculation of the dynamic benchmark price.

[0071] Figure 4 This is a schematic diagram of the entire process of preventing collusion and blockchain-based evidence storage. Detailed Implementation

[0072] Example 1: Standardized Commodity Procurement (A4 Printing Paper) (This example is used to specifically illustrate the technical features of claims 1, 3, and 6)

[0073] 1. Scenarios and technical issues

[0074] Background: A large enterprise is purchasing 10,000 boxes of A4 printing paper (a highly standardized commodity) in a single transaction through an intelligent bidding platform. Five suppliers (S1 to S5) are participating in the bidding.

[0075] Core technical issues:

[0076] How can we systematically screen out suppliers with high performance risks in the vicious competition that can easily be caused by the "lowest price wins" rule?

[0077] How can we eliminate any possibility of collusion during the bidding process?

[0078] How can we subtly optimize the price distribution through systemic intervention, guiding towards a healthier and more stable fair market price?

[0079] 2. System Execution Process (Detailed Data Flow and Algorithm Intervention)

[0080] Step 1: Quotation Reception and Information Isolation Take Effect

[0081] The system received the following final quotes: S1: 50 yuan / box, S2: 52 yuan / box, S3: 40 yuan / box (extremely low price), S4: 51 yuan / box, S5: 49 yuan / box.

[0082] The full-domain zero-knowledge information isolation strategy will be activated immediately:

[0083] Supplier S1's front-end interface only displays: "Your current quote: 50 yuan / box". There are no other supplier identifiers, no rankings, no total number of employees, and no benchmark price.

[0084] The demand side interface only displays: "Current valid quote quantity: 5". All supplier identities and quote details are completely hidden.

[0085] Step 2: Calculate the dynamic benchmark price and identify risks

[0086] The system calculates the trimmed average: After removing the highest 10% (52 yuan) and lowest 10% (40 yuan) bids, the dynamic benchmark price = (50 + 51 + 49) / 3 = 50 yuan / box. (For example, when there are 7 valid bids, the removal quantity is 7 × 10% = 0.7, rounded up to 1, meaning the average is calculated after removing 1 lowest bid and 1 highest bid; when there are 14 valid bids, the removal quantity is 1.4, rounded up to 2, meaning the average is calculated after removing 2 lowest bids and 2 highest bids.)

[0087] The S3 quoted at 40 yuan was marked as a "high-risk quote" by the system, and its historical performance record showed that it had three delayed deliveries.

[0088] Step 3: Generate and send personalized fuzzy guides

[0089] The system executes an asymmetric guidance strategy based on a dynamic benchmark price.

[0090] Send only one private guidance message to S2: "Your bid deviates from the reasonable range. We suggest adjusting it to the range of 49-51 yuan to significantly increase the probability of winning the bid."

[0091] For S1, S4, and S5 (quotes at or near a reasonable range), no guidance information is sent, and their interfaces remain unchanged.

[0092] For S3 (offering too low a price), do not send any guiding information to prevent it from using this signal for gambling.

[0093] After receiving the instruction, S2 adjusted the price from 52 yuan to 50 yuan.

[0094] Step 4: Bidding and Order Allocation Based on Dynamic Benchmark Price

[0095] At the end of the bidding period, the system sorts bids by the absolute value of the difference between the quoted price and the dynamic benchmark price: S1:|50-50|=0; S2:|50-50|=0; S5:|49-50|=1; S4:|51-50|=1; S3:|40-50|=10.

[0096] The sorting result shows that S1 and S2 are tied for first place. According to the preset "order splitting" rule, the system automatically splits the 10,000-box order into two sub-orders of 5,000 boxes each, and assigns them to S1 and S2 respectively.

[0097] 3. Achieved technical effects

[0098] System stability and risk control: The system automatically avoids S3 issues caused by extreme pricing and poor historical performance through average price benchmark rules. By evenly distributing orders, supply chain dependence is diversified, resulting in a qualitative leap in system stability.

[0099] True "high quality at a low price": The winning bid of 50 yuan is a fair market price, guaranteeing the supplier's reasonable profit and enabling them to provide high-quality products and fulfill the contract on time. Complete defense against collusion: The entire process is "completely invisible," allowing suppliers to quote prices as if in an independent vacuum, fundamentally eliminating the information basis for collusion. Personalized, ambiguous guidance prevents signals from being exploited collaboratively.

[0100] Intelligent market optimization: Through precise "surgical" guidance, the system intervenes only in S2 and quietly optimizes the entire price distribution to the most concentrated and healthy state, demonstrating the intelligence of the algorithm.

[0101] Example 2: Procurement of Non-Standardized Services (Software Development)

[0102] 1. Scenario and Technical Issues (This embodiment is used to specifically illustrate the technical features of claims 1, 3, 4, and 5, and to demonstrate small sample fault tolerance and the new supplier credit scoring mechanism)

[0103] Background: A technology company (the client) needs to purchase a customized "intelligent warehouse management system" development service. This project is technically complex and involves multiple evaluation dimensions, making it a typical non-standard project.

[0104] Core technical issues:

[0105] How can we ensure the robustness of the benchmark price and avoid distortion by extreme pricing when dealing with a small sample size (few suppliers)?

[0106] How can we provide a level playing field for new supplier C (who lacks historical data on the platform) and break down barriers to entry?

[0107] In a market environment where supply exceeds demand, how can algorithms automatically adjust decision-making priorities to ensure system stability and delivery quality?

[0108] 2. System Execution Process (Detailed Data Flow and Algorithm Intervention)

[0109] Step 1: Quotation Receiving and Preprocessing

[0110] The system received final quotes from three suppliers: Company A: 120,000 yuan; Company B: 110,000 yuan; Company C (new supplier): 130,000 yuan.

[0111] The proactive anti-collusion module performed real-time verification and found no related behavior, so all bids were deemed valid.

[0112] Step 2: Adaptive calculation of dynamic benchmark price (core algorithm involved)

[0113] The system detected 3 valid bids, triggering the "small sample tolerance mechanism".

[0114] Internal logic judgment: The quantity 3 < the first threshold (5), but ≥ the second threshold (3). Therefore, the trimmed mean is automatically abandoned and the "median" algorithm is switched. In the small sample fault tolerance mechanism, the specific values ​​of the first threshold and the second threshold can be configured according to the actual application scenario. The setting principle is that the first threshold needs to ensure that there is a sufficient sample size to calculate a stable trimmed mean, and it is usually recommended to take a value of not less than 4; the second threshold is used to distinguish extremely small samples, and it is usually recommended to take a value of not less than 2. Those skilled in the art will understand that, for example, setting the first threshold to 4, 5 or 6, and setting the second threshold to 2 or 3, can achieve the same fault tolerance purpose, and these transformations all fall within the protection scope of this invention.

[0115] Calculation process: The bids are sorted as [11, 12, 13]. Dynamic benchmark price (median) = 120,000 yuan.

[0116] (Comparison: If the arithmetic mean is (11+12+13) / 3 = 120,000 yuan, although the result is the same, the median algorithm is theoretically more resistant to interference in this scenario. If C's price is 200,000 yuan, the arithmetic mean is approximately 143,000 yuan, which has been distorted, while the median remains at 120,000 yuan, demonstrating its superiority.)

[0117] Step 3: Dynamically adjust decision weights (adaptive strategy)

[0118] The system also detected that the recent supply and demand index for this type of software development service is >1.2 (supply exceeds demand).

[0119] According to preset rules, the "price weight coefficient" in the comprehensive score is automatically lowered from the baseline value of 0.6 to 0.5, while the "credit and performance weight" is increased from 0.4 to 0.5. This shifts the decision-making focus from "price" to "quality and reliability." As a preferred implementation, the real-time market supply and demand index is specifically defined as the ratio of the total supply from participating suppliers to the aggregated order demand from demanders.

[0120] Step 4: Comprehensive Score Calculation (Data Fusion Decision)

[0121] The system obtains real-time data from the credit profiling module:

[0122] Company A: Credit score of 95 (Excellent), historical performance rate of 98% over the past 30 months.

[0123] Company B: Credit score 80 (good), historical performance rate of 90% over the past 30 months.

[0124] Company C: Credit score of 80 (the base score assigned by the system to new suppliers), slightly higher based on its qualification documents (CMMI3 certification); fulfillment rate of 85% (assessed based on its third-party cooperation certification).

[0125] Overall score calculation (price weight 0.5, credit and performance weights 0.25 each): Company A:

[0126] Price points = (1 - |12 - 12| / 12) * 100 * 0.5 = 50

[0127] Credit score=95*0.25=23.75

[0128] Performance score = 98 * 0.25 = 24.5

[0129] Total score = 50 + 23.75 + 24.5 = 98.25

[0130] Company B:

[0131] Price score = (1 - |11 - 12| / 12) * 100 * 0.5 ≈ 45.83

[0132] Credit score=80*0.25=20

[0133] Performance score = 90 * 0.25 = 22.5

[0134] Total score = 45.83 + 20 + 22.5 = 88.33

[0135] Company C:

[0136] Price score = (1 - |13 - 12| / 12) * 100 * 0.5 ≈ 45.83

[0137] Credit score=80*0.25=20

[0138] Performance share = 85 * 0.25 = 21.25

[0139] Total score = 45.83 + 20 + 21.25 = 87.08

[0140] Step 5: Awarding and Information Separation

[0141] The scores are sorted from highest to lowest based on overall evaluation: A (98.25) > B (88.33) > C (87.08). The system determines that Company A wins the bid.

[0142] Information isolation in effect:

[0143] The demand side's interface only displays: "Winning Bidder: Company A, Winning Bid Price: 120,000 RMB, System Evaluation of Performance Reliability: Excellent." All information about Companies B and C is completely hidden. Supplier A's interface displays: "Congratulations on winning the bid!" Suppliers B and C's interfaces respectively display: "Unfortunately, you did not win this bid; your overall competitiveness needs improvement." All suppliers are unaware of the existence, bids, and specific rankings of other participants.

[0144] 3. Achieved technical effects

[0145] Small sample tolerance: By adaptively switching to the median algorithm, the system ensures that the dynamic benchmark price remains robust and fair even when facing a small number of suppliers, providing a reliable benchmark for subsequent decision-making.

[0146] The effect of new supplier access: Through the credit scoring mechanism, new supplier C gained the opportunity to compete with established suppliers. Its total score was only slightly different from that of company B, successfully breaking down data barriers.

[0147] Dynamic decision-making and quality assurance effects: By dynamically adjusting the price weight, the system automatically prioritizes Company A, which has the best credit and performance record, in scenarios where supply exceeds demand, even though its price is not the lowest. This directly improves the certainty of on-time and high-quality project delivery and fundamentally optimizes system stability.

[0148] Information isolation and anti-collusion effect: The "two-way stealth" throughout the process makes collusion technically impossible, ensuring the fairness of the bidding process.

[0149] Example 3: Adaptive Benchmark Price Calculation and Decision-Making in Small Sample Scenarios (This example is used to specifically illustrate the technical features of claims 3 and 5, focusing on demonstrating the application of the median algorithm and automatic weight adjustment when the number of bids = 3)

[0150] A company is purchasing a batch of special-specification components. Only three suppliers (D, E, and F) have the capability to supply them and are participating in the bidding.

[0151] Step 1: Receive quotes. The quotes are as follows: D: 1000 yuan / piece, E: 1200 yuan / piece, F: 3500 yuan / piece (extreme quotes).

[0152] Step 2: Adaptively calculate the dynamic benchmark price. The system detects that the number of valid bids is 3, which is less than the first threshold (5) but greater than or equal to the second threshold (3), so it automatically switches to the median algorithm. The bids are sorted as: 1000, 1200, 3500. The dynamic benchmark price (median) = 1200 yuan.

[0153] Step 3: Decision Making. This component is a non-standardized product, requiring a comprehensive score calculation. Due to the limited number of suppliers triggering a small sample strategy, the system automatically reduced the price weight from the usual 0.6 to 0.4, and increased the credit and performance weights from 0.2 to 0.3 each. After calculation, supplier E was awarded the bid because its price was closest to the benchmark price and it demonstrated good credit and performance.

[0154] Results: Through adaptive algorithms, the system still calculates a robust dynamic benchmark price even in unfavorable scenarios such as a small number of suppliers and extreme bidding. By adjusting decision weights, it effectively eliminates the interference of irrational bidding and ensures the reliability of the bidding results.

[0155] Example 4: AI-Assisted On-Chain Arbitration (Demonstrating Automated Dispute Resolution) (This example is used to specifically illustrate the technical features of claim 8 and demonstrate the on-chain arbitration and smart contract automatic execution process)

[0156] 1. Scenarios and technical issues

[0157] Background: In a derivative scenario of Example 1, the supplier, "Precision Components Company," won a bid for 5,000 boxes of packaging. Due to a production line malfunction, the delivery was delayed by 5 days, causing operational losses to the customer, "Speedy E-commerce." A dispute arose between the two parties regarding the amount of compensation.

[0158] The core technical issue is: how to resolve such performance disputes in a low-cost, efficient, and credible manner, avoiding the pain points of time-consuming and laborious traditional judicial litigation?

[0159] 2. System Execution Process (Detailed Data Flow and Algorithm Intervention)

[0160] Step 1: Dispute Triggering and Data Preparation

[0161] The party in need initiates an arbitration application on the platform.

[0162] The system automatically retrieves and verifies all stored hashes related to the order from the blockchain: Hash_A: Bidding rules (including default clauses).

[0163] Hash_B: All bids and winning results.

[0164] Hash_C: The funds and margin locked in the smart contract.

[0165] Hash_D: The electronic contract signed by both parties (with a specified delivery date).

[0166] Step 2: AI-assisted analysis

[0167] The arbitrator (an industry expert invited by the platform) authorized the activation of the AI ​​analysis model.

[0168] The AI ​​model acquires verified, trusted data from the blockchain and performs the following analysis:

[0169] Liability determination: By comparing the delivery date stipulated in the contract with the actual logistics data, it was confirmed that the "precision parts company" delayed delivery, and the responsibility was clear.

[0170] Loss Assessment: The model is connected to the e-commerce platform API (with authorization from the demand side) to analyze the sales losses and warehousing cost fluctuations of the product on "Suda E-commerce" during the delay period, and to make a comprehensive calculation in conjunction with the liquidated damages clauses stipulated in the contract.

[0171] Report generation: The AI ​​model generates a "Recommendation Report on Liability and Loss Analysis of Contract Performance Disputes", which clearly identifies the responsible party and provides a suggested range of compensation amount (e.g., 3,000-4,000 yuan) and detailed calculation basis.

[0172] Step 3: Arbitration Award and Smart Contract Execution

[0173] The arbitrator, referring to the professional advice generated by AI and combining it with his own industry experience, made a final ruling: "Precision Components Company" is in breach of contract and must pay "Suda E-commerce" a penalty of 3,500 yuan.

[0174] The ruling was digitized and a hash Hash_E was generated and written to the blockchain.

[0175] The smart contract was automatically triggered: 3,500 yuan was transferred from the previously frozen deposit to the "Suda E-commerce" account, and the remaining payment was unfrozen and given to the "Precision Components Company".

[0176] The entire transfer process is completed within 2 minutes, fully automated, and requires no further confirmation from either party.

[0177] 3. Achieved technical effects

[0178] High efficiency and low cost: From initiating arbitration to enforcing compensation, the entire process can be completed within hours, improving efficiency by more than 90% and reducing costs by more than 80% compared to traditional judicial procedures.

[0179] The arbitration is fair and credible: based on immutable evidence throughout the entire process on the blockchain, it eliminates any room for disputes. AI assistance makes loss assessment more scientific and objective, enhancing the professionalism and credibility of the ruling.

[0180] Execution is absolutely reliable: Through automatic execution via smart contracts, the risk of the losing party delaying or refusing to pay is avoided, achieving "execution of the judgment as soon as it is issued".

[0181] Enhanced Ecosystem Trust: This mechanism builds a strong foundation of trust for the platform, attracting more high-quality suppliers and demanders who prioritize transaction security to join, thus creating a network effect.

[0182] Example 5: Anti-collusion behavior identification and handling (demonstrating proactive backend defense) (This example is used to specifically illustrate the technical features of claim 7, and fully present the identification and handling process of the "probing pricing behavior pattern")

[0183] 1. Scenarios and technical issues

[0184] Background: In a large-scale, long-term procurement bidding process for industrial standard parts, there is a potential organized supplier alliance attempting to test and break the system's bidding rules.

[0185] Core technical issue: When the system is completely invisible on the front end and colluders cannot directly identify their accomplices, how can it proactively identify, verify, and combat coordinated bidding behavior achieved through sophisticated probing?

[0186] 2. System Execution Process (Detailed Data Flow and Algorithm Intervention)

[0187] Step 1: Backend Behavioral Data Collection and Anonymous Identification Management

[0188] The system maintains a random anonymous identifier (such as Supplier_7a3F, Supplier_k8qT) for each supplier in the background. This identifier is associated with the real identity and is reset periodically every 24 hours.

[0189] The proactive anti-collusion module continuously collects device fingerprints (browser Canvas fingerprints, IP address ranges), network environment data, and pricing behavior sequences accurate to milliseconds.

[0190] Step 2: Multidimensional Feature Analysis and Anomaly Pattern Recognition

[0191] During three consecutive bidding rounds, the module detected:

[0192] The price difference between Supplier_7a3F and Supplier_k8qT has remained consistently within 0.5%.

[0193] The timestamps of the two quotes are displayed at fixed intervals across multiple rounds (e.g., a difference of 2 seconds between each quote).

[0194] The device fingerprint information shows that both devices had previously logged in using the same IP address range.

[0195] The time-series behavioral analysis model determined that the behavioral sequence belonged to the "tentative collaborative bidding" pattern, with a confidence level as high as 92%.

[0196] Step 3: Proactive Intervention and Tiered Punishment

[0197] The system immediately triggered a "non-periodic anonymous identifier reset". Supplier_7a3F and Supplier_k8qT were re-marked as Supplier_p9mZ and Supplier_r2xN in the background. This foreground operation was undetectable.

[0198] At the same time, the system initiates tiered processing according to the rules:

[0199] Level 1 penalty: Immediately freeze the bidding permissions of both parties for 24 hours.

[0200] Level 2 penalty: 10 points will be deducted from their credit score.

[0201] The penalty decision and the entire chain of evidence for behavioral analysis are hashed and written to the blockchain for storage.

[0202] Step 4: Results Feedback

[0203] After logging in, the penalized supplier's interface displays: "Your account has been temporarily frozen due to detected abnormal bidding behavior. Please contact customer service for details." The system does not disclose any information about the monitoring algorithm, related parties, or other suppliers.

[0204] 3. Achieved technical effects

[0205] Proactive defense: The system no longer passively accepts quotes, but takes the initiative through background behavior analysis to dismantle collusion before it causes actual damage.

[0206] Precision strike: Based on high-dimensional features and time series models, the system can effectively distinguish between normal competition and malicious collusion, avoiding accidental damage and achieving precise strikes.

[0207] Deterrence and Ecosystem Purification: Timely penalties and credit score deductions create a strong deterrent, continuously purifying the platform's competitive ecosystem and fundamentally suppressing the motivation for collusion.

[0208] Complete chain of evidence: Blockchain-based evidence storage ensures the fairness and immutability of penalty decisions, providing judicially credible evidence for potential disputes.

[0209] In the blockchain evidence storage step, the "blockchain" mentioned is preferably a consortium blockchain architecture to ensure the confidentiality of data during the bidding process and the verifiability after the bid is awarded.

[0210] The specific implementation method is as follows:

[0211] Alliance Construction: The alliance blockchain is maintained by this bidding platform, one or more judicial appraisal centers, and relevant industry associations as consensus nodes. This multi-centralized governance model ensures data immutability while avoiding the risk of control by a single entity. Confidentiality Mechanism: During the bidding process, the hash values ​​of all key data to be verified (such as bids and dynamic benchmark prices) are written to the alliance blockchain. This data itself is stored confidentially using encryption technology or in an off-chain database. Only after the winning bidder is determined is the system authorized to decrypt or publicly disclose its mapping relationship to the alliance blockchain nodes for verification by the winning bidder, the requesting party, or the arbitration party. Trusted Verification: Any third party can verify the fairness and data integrity of the entire bidding process by querying the timestamped hash value records stored on the alliance blockchain and comparing them with the final bidding results data published by the platform.

[0212] Those skilled in the art will understand that, in addition to consortium blockchains, the blockchain-based evidence storage can also be implemented on public or private blockchains. However, consortium blockchains achieve the best balance between "process confidentiality," "judicial credibility," and "execution efficiency," making them the preferred solution of this invention. Any solution that uses blockchain technology to implement process data storage and post-verification falls within the protection scope of this invention.

[0213] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart bidding method based on a dynamic benchmark price, executed by one or more processors of a data processing system, comprising the following steps: Receive quotations from multiple suppliers for a specific subject matter, which includes a dynamic order pool aggregated from transaction requests from multiple demanders; Calculate the dynamic benchmark price for all valid bids; Based on the dynamic benchmark price, a stability-first decision rule is applied to determine the winning bidder; wherein, the stability-first decision rule is: Determine whether the subject matter is a standardized subject matter based on its attribute information; If the subject matter is standardized, the suppliers will be ranked from smallest to largest based on the absolute value of the difference between their bids and the dynamic benchmark price, and the best bidder will be selected. If the subject matter is non-standardized, the bid will be awarded based on the comprehensive score from high to low. The comprehensive score is calculated by weighting at least the price sub-score, credit sub-score, and performance capability sub-score, and the price sub-score is calculated based on the dynamic benchmark price. Standardized targets refer to targets whose core attributes can be quantified and are compatible with industry standards, while non-standardized targets refer to targets whose core attributes need to be customized or have no unified standards.

2. The method according to claim 1, characterized in that, The dynamic benchmark price is the arithmetic mean, trimmed mean, or median.

3. The method according to claim 2, characterized in that, The calculation process of the dynamic benchmark price includes a small sample fault tolerance mechanism: When the number of valid bids is greater than or equal to the first threshold, the trimmed average is used as the dynamic benchmark price; When the number of valid bids is less than the first threshold but greater than or equal to the second threshold, the median is used as the dynamic benchmark price. When the number of valid bids is less than the second threshold, the arithmetic mean is used as the dynamic benchmark price, and the system's decision weight adjustment strategy is triggered. Wherein, the first threshold is greater than the second threshold.

4. The method according to claim 1, characterized in that, The price sub-score is calculated as follows: Price Score = MAX(0, (1-|the supplier's quoted price - dynamic benchmark price| / dynamic benchmark price)) × 100 × price weight coefficient.

5. The method according to claim 4, characterized in that, The price weighting coefficient is a dynamic parameter, and its value is dynamically determined by the system based on at least one of the following factors: the standardization degree of the target, the real-time market supply and demand relationship, and the supplier's credit rating. The real-time market supply and demand index is used to quantify the real-time market competition situation. It is calculated based on the ratio of the total supply capacity of the supply side to the total demand of the demand side. Specifically, when the real-time market supply and demand index is higher than a preset first threshold, the system automatically lowers the price weighting coefficient; when the supplier's credit score is lower than a preset second threshold, the system automatically lowers the price weighting coefficient in the supplier's comprehensive score calculation.

6. The method according to claim 1, characterized in that, The method also includes an intelligent guidance step: Based on the dynamic benchmark price, the system generates personalized and asymmetric fuzzy guidance information for each supplier and pushes it through their private interface. The guidance strategy is asymmetric. The system only sends quantitative guidance information containing suggestions to adjust the price range to suppliers whose quotations are higher than the reasonable range, while it does not send any or only sends neutral prompts to suppliers whose quotations are too low. The reasonable range is a price range determined by a preset percentage, which fluctuates up or down based on the dynamic benchmark price. The preset percentage range is 5% to 15%, and the specific value is determined according to the type of the target asset.

7. The method according to claim 1, characterized in that, The method also includes an active anti-collision step: A random, anonymous identifier is generated for each supplier and periodically reset in the system backend; Collaborative analysis is performed based on multidimensional features of device fingerprints, network environment, and pricing behavior sequences; When a tentative bidding behavior pattern is detected, the random anonymous identifier is immediately reset non-periodicly. The tentative bidding behavior pattern refers to a characteristic sequence identified by the system's background time-series behavior analysis model, in which at least two suppliers have a stable bid difference within ±1% in two or more consecutive bidding rounds, and the bid time interval shows a fixed pattern. A time-series behavior analysis model is used to identify collusion patterns, and the identified abnormal behaviors are processed in a tiered manner.

8. The method according to claim 1 or 7, characterized in that, The method also includes information isolation and rule-based evidence preservation steps: A) Information Isolation: Implement a role-based, zero-knowledge information isolation strategy across the entire domain, and implement differentiated information access controls for suppliers, demanders, non-participants, visitors, and system administrators to ensure that no role can obtain information that can be used for collusion; B) Rule storage: The bidding decision rules, all valid bids, dynamic benchmark price and winning bid results are generated into hash values ​​and written into the blockchain, and made available for public verification after the winning bidder is determined.

9. An intelligent bidding system for implementing the method of any one of claims 1 to 8, comprising a modular architecture design, including: One or more processors; as well as One or more computer-readable storage media for storing computer instructions executable by the processor; The core functional modules of the system communicate and call each other through predefined interfaces; these interfaces are configured to connect to external data sources and extension plugins that conform to their specifications, so as to realize the system's functional expansion, technical upgrades and update maintenance; When executed by the processor, the instructions configure the system to implement the method as described in any one of claims 1 to 8.

10. The system according to claim 9, characterized in that, The core functional modules include: The adaptive decision engine module is configured to perform dynamic benchmark price calculation, adaptive weight coefficient adjustment, and intelligent guidance strategy generation. The credit profiling and ecosystem management module is configured to integrate multi-source data to generate credit assessments for suppliers and demanders, and to assign a preset credit score base to new suppliers that lack historical data. The strict information isolation control module is configured to implement a role-based, global zero-knowledge information isolation strategy, and to implement differentiated information access control for suppliers, demanders, non-participants, visitors, and system administrators. The proactive anti-collusion module is configured to perform dynamic anonymous identifier management, collusion intent identification, and behavior sequence analysis, judgment, and handling. The blockchain evidence storage and smart contract module is configured to handle the entire process of data on-chain, deploy and manage smart contracts, and initiate an AI-assisted on-chain arbitration process in the event of a dispute.

Images